New Orleans—A recent study presented by Raisa Pompe and colleagues during a session at the AUA 2015 Annual Meeting indicated that multiparametric classifiers have a high potential to improve the predictive accuracy of distinguishing between indolent and aggressive forms of prostate cancer.
Better prognostic markers are needed for prostate cancer, according to the researchers, and while single markers are not sufficient for routine application, first generation commercial multiparametric test assays have shown promising results.
The researchers analyzed more than 12,000 prostate cancer patients using long-term follow-up data of a tissue microarray containing 0.6 mm tissue cores. More than 80 different antibodies and DNA probes by immunohistochemistry and fluorescence in situ hybridization were analyzed.
Markers with strong and independent prognostic value in univariate and multivariate analyses were combined to obtain panels of markers that produce additional prognostic information beyond the standard measures, which include preoperative prostate-specific antigen, Gleason grade, tumor stage, and nodal stage.
Via a receiver operating characteristic analysis, a score of zero to 10 was produced based on the number and type of adverse features per tumor.
The researchers found the following markers had strong independent prognostic relevance:
- 5q21 and PTEN
- Overexpression of RBM3
- Nuclear accumulation of p53
All of them indicated a dominant-negative TP53 gene mutation.
A total of 4159 cancers had a score of zero and included none of these alterations. The worst prognosis was found for 570 tumors with scores of eight or more of the alterations. An intermediate prognosis with decreasing time to recurrence was found for 2365 cancers with scores of one or two, 1499 cancers with scores of three or four, and 1111 cancers with scores of five through seven (P<.0001).
In multivariate analyses, the 5 gene classifications provided prognostic information independent from the standard parameters (P<.0001).
The researchers concluded that “analyses of very large sets of patient tissues are instrumental to identify the best candidates for such molecular marker panels.”
Source: Pompe R, Kraft S, Gild P, et al. A multiparametric molecular classifier for improved prediction of prostate cancer prognosis. Abstract of a presentation at the American Urological Association 2015 Annual Meeting, New Orleans, Louisiana, May 17, 2015.